Unverified50% confidenceBenchmarkExact time
根据Artificial Analysis数据，Fable完成单个任务约用33k token，Opus 5约用37k token
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/23/2026
First Seen
Valid until: 11/21/2026
Sources
Related Entities
Related Claims
Unverified在Gemma 4 12B模型的Token生成测试中，Spark达26.4 tokens/s、Halo 24.6 tokens/s、M4 Pro最快33.8 tokens/s73% similarUnverified父子分块策略用小块（128~256 Token）做精准检索，命中后返回其所属大块（512~1024 Token）作为上下文71% similarUnverifiedMiniMax M3 每个 token 推理时约激活 23B 参数63% similarUnverified根据Chinchilla缩放定律，一个30B模型的最优训练数据量约为600B-1T tokens62% similarUnverifiedIn the e-commerce website test, Opus 4.8 consumed 4.898 million output tokens while Fable 5 used only 588K tokens62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/788642API
curl https://kongchang.com/api/v1/knowledge/claims/788642MCP
get_claim(id=788642)